korowa commented on issue #14238: URL: https://github.com/apache/datafusion/issues/14238#issuecomment-2614501666
As a further step, I'd like to also try it for HashJoin because of this point > join can either be smaller or **larger** than the target batch size Now hash join produces batches with less than or equal to batch_size rows, and in cases when the number of output rows is greater than the number of input rows (some non-unique keys on build side) it may end up with a sequence of batches sized as ``` batch_size -> (0.1 * batch_size) -> batch_size -> (0.1 * batch_size) ``` where small batches are the result of key non-uinqueness. This sequence won't be compacted by CoalesceBatchesExec, due to full-sized batches, but having internal join buffer for the output may help with better output batch size alignment -> less number of batches. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org For additional commands, e-mail: github-h...@datafusion.apache.org
